Project Overview
This project was initiated by the Inventory and Marketing teams of Kalbe Nutritionals to tackle two primary objectives:
- Sales Prediction: Forecast daily sales quantities to help the inventory team maintain optimal stock levels.
- Customer Segmentation: Cluster customers based on key attributes to help the marketing team provide personalized promotions and targeted sales treatment.
Tools Used
- Python: For data manipulation, analysis, and modeling.
- Jupyter Notebook: For model development.
- PostgreSQL: As the database to store and query data.
- DBeaver: For querying and database management.
- Tableau Public: For creating interactive visual dashboards.
Datasets
Four main datasets were used in this project:
- customer.csv: Contains information about customers, including their age gender, marital status, and income.
- product.csv: Provides details on the products sold, such as product names and prices.
- store.csv: Contains data on store locations, store types, and store groups.
- transaction.csv: Holds transactional records with product sales data, including dates, quantities, total amounts, and store information.
Steps Taken
1. Data Ingestion
Imported datasets (customer.csv, product.csv, store.csv, transaction.csv) into PostgreSQL using DBeaver.
2. Exploratory Data Analysis (EDA)
Imported datasets (customer.csv, product.csv, store.csv, transaction.csv) into PostgreSQL using DBeaver.
2. Exploratory Data Analysis (EDA)
- Average customer age by marital status: SQL query shows average age varies slightly across marital statuses.
- Average customer age by gender: Slight difference in average age between genders.
- Top-selling store by quantity: Identified the store with the highest sales quantity.
- Best-selling product by total amount: Identified the product generating the most revenue.
3. Data Visualization (Tableau)
Created four worksheets and a dashboard for visualizing:
- Monthly sales quantity trends.
- Daily total sales amount.
- Sales by product.
- Sales by store.
4. Predictive Modeling (Sales Prediction)
Built a Time Series Regression Model using Python to predict daily sales quantities.
- Model trained on transaction data.
- Helps the inventory team maintain optimal stock levels based on predicted daily demand.
5. Customer Segmentation (Clustering)
- Applied K-Means Clustering to segment customers.
- Resulted in 5 distinct customer segments for personalized marketing strategies.
Results
- Sales Prediction: The regression model predicts daily sales quantities, aiding the inventory team in stock management.
- Customer Segmentation: Segments based on demographic data help the marketing team create targeted campaigns.
More About This Project
You can visit the repository of this project by click here.